Jump to:
In early March, while the local and international public agenda was dominated by the political and economic stakes of the war in Iran, Beijing published one of the most interesting economic documents in recent times. The 15th Five-Year Plan (2026–2030) is a 141-page blueprint that formalizes the transition from “High-Quality Growth” to “Total Technological Sovereignty.”
In this article, we dissect the Chinese strategy in detail, we also look at the American route and use them as a mirror for the Romanian reality. Our message is one of extreme urgency: while China surgically programs the year 2030 and does not shy away from building long-term plans precisely in the era of rapidly advancing technology, and the US relies on its companies to gain global market share, Romania risks remaining a spectator.
Analysis: Digital Core, AI as a national operating system
The 15th Five-Year Plan marks the official transition of the “AI Plus” (AI+) initiative from the pilot program stage to a national mandate. The frequency of mentions of Artificial Intelligence (over 50 times in the document) underlines its role as the main driver of what Beijing calls “New Quality Production Forces.” The terms may bring a bitter smile to your face, rooted in communist-era language, but the Chinese approach is as serious as can be.
AI penetration and economic value
The plan sets two of the most ambitious technological targets ever seen in a state-run economy:
- Target for 2027: A 70% AI penetration rate across all “key sectors” (manufacturing, energy, logistics, and government).
- Target for 2030: A 90% AI penetration rate, making “AI-native” operations the default standard of the Chinese economy.
At the valuation level, the entire AI industry is forecast to exceed the value of 10 trillion yuan (~$1.38 trillion) by 2030.
Pillars of digital technology
Humanoid Robotics is designated as a “Key Pillar Industry.” The plan explicitly calls for doubling the production of humanoid robots in five years. The goal is “embodied AI,” which are robots that not only follow instructions but can navigate autonomously in complex industrial and domestic environments.
Brain-Computer Interfaces (BCIs): The plan moves BCIs from the “experimental” category to that of “emerging strategic industries.” Areas of interest include medical rehabilitation, “cognitive optimization” for the highly skilled workforce, and “neuro-command” systems for defense.
Quantum Networks in Space: To overcome the bottleneck created by the “Great Firewall” and secure communications against eavesdropping, the 15th Five-Year Plan commits to creating a global-scale quantum communications network. This includes launching state-of-the-art quantum satellites to create an unbreakable “Quantum Internet” by 2030.
Infrastructure as evolution
To support a 90% AI penetration rate, China is doubling its investment in computing infrastructure. The plan outlines a massive expansion of the “National Unified Computing Power Network.”
EFLOPS (Exa-Floating Point Operations Per Second): It represents a billion calculations per second; it is the unit of measurement for a nation’s hardware “muscle.” The 1200 EFLOPS target represents the raw infrastructure needed for AI to run 90% of China’s economic processes in real time.
“Extraordinary Measures” for Sovereignty
Beijing has acknowledged that a digital economy is only as strong as its physical foundations. The plan uses the phrase “extraordinary measures” to describe its approach to two critical areas: semiconductors and rare earths.
Self-sufficiency in Semiconductors
The plan recognizes that traditional market mechanisms have failed to catch up with the West in high-performance lithography. The “extraordinary measures” begin with the “Whole Nation” Fund: a massive mobilization of state capital to circumvent US export controls. This measure is complemented by Vertical Integration, the obligation of state-owned enterprises and large tech firms to use a minimum percentage (probably 70-80%) of domestic chips for all AI-related infrastructure by 2027. The latter concludes the triad of extraordinary measures. Development of general-purpose quantum computers and proprietary high-performance AI chip architectures that do not depend on Western-controlled technologies such as ARM from the UK/Japan or x86 from the US.
Rare Earths and Strategic Resources
China is shifting from a “supplier” to a “strategic user” of critical minerals. As such, it needs export control systems by refining its licensing system to differentiate approvals based on “diplomatic alignment” with Beijing. The plan also aims to control 95% of global heavy rare earths processing capacity, ensuring that even if other nations mine the minerals, they must still be sent to China for refining.
Energy and Frontier Sciences
The plan treats the energy transition as a “computational problem.” High-density AI requires massive power, leading to a new focus on nuclear fusion. The document sets a strict timetable for the commercialization of controlled nuclear fusion and proposes completing the prototype phase by 2030, targeting a pilot project of a “fusion-powered data center” in the early 2030s.
In addition, it details a relatively little-discussed topic in Europe or in Bucharest: the Low Altitude Economy. We are talking about the accelerated growth of drone-based logistics and eVTOL (electric vertical takeoff and landing) transport and, therefore, the use of airspace below 1000m for transport and services to reduce urban congestion and integrate AI-driven logistics into the “last mile” of deliveries.
Conclusions of the document
The 15th Five-Year Plan is essentially a blueprint for a “Fortress China.” It is designed to completely decouple the Chinese economy from Western technological dependencies, while making the world more dependent on Chinese-processed resources and standardized hardware through Chinese AI.
The strategy details how Beijing aims to build and export complete systems: computing infrastructure, cloud services, digital platforms and governance frameworks. Initiatives such as the Digital Silk Road or the development of offshore data centers point to a clear direction: technological capacity is projected beyond borders, especially towards emerging economies.
This approach comes with a coherent economic and political model. Digital infrastructure is supported by dedicated financing mechanisms, low-cost energy and export credit instruments. In parallel, China is expanding its influence through standards and regulatory models, especially in its relationship with the Global South. In practice, this means that countries adopting these systems not only take over technology, but also structural dependencies and operating rules.
The direction is also visible in the way economic success is beginning to be measured. The focus is shifting to the revenues generated from globally operated digital infrastructures and services. In this context, competition is not limited to the development of applications or products, but includes control of the infrastructure, data flows and standards that define the use of artificial intelligence at scale.
By focusing on local “total penetration” and exporting a complex system, China is not just trying to build better AI, but is trying to develop an economy where AI is the foundation, making traditional economic indicators (such as GDP growth) less relevant than “Total Factor Productivity” (the ability to produce significantly more with a shrinking workforce).
In conclusion, if the 14th Five-Year Plan was about “Survival,” the 15th is about “Supremacy.” The document signals that China sees the “Great Transition” to an AI-driven world as inevitable and intends to be the architect of its standards.
Finally, a word of warning. There are certainly many questions that we cannot miss, related to almost any document published by the authorities in Beijing. Is this document more of a plan with political communication objectives of the Communist Party or does it rest, fundamentally, on real economic substance? Is it another strategy that no one can really monitor outside of internal reporting? Can a 141-page plan guarantee innovation?
Regardless of the answer to these questions, we can take from this document an explicit intention, a direction, perhaps a vision:
China aims to lock in scarce materials, robotics and artificial intelligence, low-cost energy, and financing tools in a “state-run ecosystem” to become the next global superpower by scaling and controlling supply chains dedicated to technology.
We will see whether China’s route, which is trying to build a state-controlled, vertically integrated, and geopolitically protected AI ecosystem, will be the winner, or that of the United States, which is trying to maintain an AI ecosystem dominated by the market but strategically supported by the government and the university system.
Between these two models will likely decide the technological architecture of the global economy in the coming decades.
United States: AI as a decentralized strategic infrastructure
If the Chinese model means detailed central planning, the American one works almost the opposite: a distributed system, built through cooperation between government, industry and universities. The United States is trying to build a dominant AI ecosystem, it does not formulate its strategy in a single five-year plan, but it can be understood through several major directions established by the White House, the Department of Commerce and federal agencies.
AI as economic and military infrastructure
In recent years, Washington has simultaneously treated AI as critical economic infrastructure, a strategic national security technology, and an engine of industrial competitiveness.
The American strategy is based on mobilizing the private sector, which concentrates most of the global research and investment in AI. Companies such as OpenAI, Google, Anthropic, Microsoft, Amazon, or Meta are not just commercial players, but part of the strategic technological architecture of the United States.
Computing infrastructure and global competition
As with China, one of the major priorities of the US government is to expand the computing infrastructure needed to develop AI models. In recent years, the US has accelerated investment in AI data centers, supercomputers dedicated to training models, and the development of specialized chips for artificial intelligence.
This infrastructure is dominated by American companies such as NVIDIA, AMD, and Intel, which control a significant part of the global production chain for AI hardware. Moreover, controlling exports of advanced chips to China is a central geopolitical tool in Washington’s strategy to slow the development of technological rivals.
Research, talent and universities
Perhaps the most important structural advantage of the United States today is its research ecosystem. Universities such as MIT, Stanford, Carnegie Mellon, and Berkeley are global centers of excellence in the field of artificial intelligence and continue to attract much of the world’s talent. In the case of AI, universities are not just academic institutions, but essential components of the innovation infrastructure.
In conclusion, Washington relies on a model summarized as follows: the state sets the strategic direction, industry develops technology, and universities produce research and multiply talent.
The result is a system less predictable than Chinese central planning, but much more dynamic.
Romania: Absence of Governance as a State Strategy
China knows who is driving the AI strategy.
The United States knows who is developing the technology.
Romania does not yet know who is responsible for its future.
There are, however, signs that the first elements of AI governance are starting to emerge, especially under the impetus of the European framework. The Romanian Government recently proposed, through a memorandum, the designation of ANCOM as the national supervisory authority and single point of contact for the application of the European Regulation on Artificial Intelligence (AI Act).
These developments are important and mark the emergence of the first real AI governance layer in Romania, but they represent primarily a supervision and compliance mechanism, built for the enforcement of European legislation. This layer operates around effects, but the direction of development and priorities remain insufficiently defined. What is (still) missing is a strategic architecture: an institution or mechanism to coordinate the development, integration and national direction of artificial intelligence.
Such an absence of development direction becomes visible in the way artificial intelligence enters the economy. The impact is concentrated in a few areas where changes are rapid and direct: in industry and services, through process optimization and production automation; in public administration, through time reduction and increased transparency; in health, through the use of data in diagnosis and resource allocation.
Without explicit priorities, these developments remain fragmented and any positive impact is minimized. Therefore, integration at scale cannot arise from isolated initiatives alone, but must primarily crystallize through coordination.
In parallel, scale depends on infrastructure – but in Romania, public data is dispersed, computing capacity is limited, access between institutions is difficult. Interoperability is often the exception.
A functional minimum is clearly required: structured, accessible data; available computing capacity; usage and access rules. Without these fundamental elements, systems cannot scale.
This foundation of value generation through the integration of AI throughout the economy can currently only evolve organically, slowly, in the absence of an executive effort from the central administration. Implementation is not a spontaneous process, it is an organized process and needs coordination, standards, guidelines, and evaluation mechanisms.
In other words, Romania is starting to build AI regulation, but it does not yet have an institution responsible for AI strategy. In this context, there are three topics that we propose on the public agenda and, as in the case of the report launched in October by the Edge Institute and commissioned to the Estonian consultants from Digital Nation, we believe that everything starts with the need to clarify roles and responsibilities.
A. Non-existent Governance: Who is responsible?
Romania’s fundamental problem is not the lack of money (PNRR funds being proof to the contrary), but the lack of a process “owner”. It is therefore no wonder that the present places us in last place in the EU on the topic of integrating AI into the economy.
Source: European Commission, DESI 2024
- Leadership Vacuum: There is currently no government structure with a clear mandate, budgetary authority and technical capacity to implement a national AI strategy. However, there is the National Strategy for Artificial Intelligence (SN-IA) 2024-2027 approved by the Government in July 2024, a document that represents the main strategic framework for the adoption and integration of AI technologies, aligning with the EU objectives of the time. On the AI front, to let 2 years pass without moving at all to turn a strategy into a concrete action plan is a major blunder.
- Implementation Teams: There is no body of experts, an “AI Task Force”, to work cross-party on this strategy or to propose concrete programs, acceleration objectives and local mechanisms for monitoring the integration of AI in the administration or the economy. In the absence of stable implementation teams, any attempt at a strategy dies with the change of a minister.
B. The conflict of realities: AI vs. “The Rail File”
The Chinese ambition of 90% AI penetration and the accelerated agenda of American companies for global growth collide in Romania with the grim present of an administration that is the last in Europe or among the countries monitored by the OECD in the digitalization of public services.
- Accelerating Digitalization: We cannot talk about integrating AI into an administration that does not have interoperable databases. AI needs structured data to deliver efficiency. While China aims to centralize 80% of public interest data for AI training, Romania does not know what data lies unused in isolated databases. As such, at the local level, AI will remain a development strategy only for the private sector, without any impact on state efficiency.
- Analog Bureaucracy: Romania needs a leap forward. We can no longer go through “classical digitalization” for 10 years and then think about AI. We need to adopt forward thinking strategies, directly integrating automated data processing solutions. Today.
C. Time horizon: 5 years vs. 1 year
This is perhaps the most painful comparison.
- The Chinese Model: 5-year planning, with rigorous intermediate stages. Of course, in today’s speed of technological change, a 5-year plan can be questionable as a roadmap, but not if you coherently detail principles, directions, and objectives.
- Romanian Model: 1-year planning now dictated by government rotations or immediate political survival and between 2028 and 2032 probably by electoral cycles.
National Consensus is the key concept. “Mega-future” projects, such as AI, cannot be right-wing or left-wing. They require a national consensus that breaks the local custom of “cancelling the predecessor’s projects.” Just as the country’s defense strategy does not change with every change of Defense Minister.
Without a cross-party pact on technology and AI, Romania will remain, on the one hand, a market for algorithms written by others, with economic costs that are difficult to understand today, and on the other hand, our country will become more vulnerable in terms of national security.
What Edge Institute proposes for the public agenda
For Romania not to miss another train of the future, we must first recognize, we believe, that the current pace of AI implementation condemns us to irrelevance. It is a necessary step to understand the risk of doing too little, too slowly, too late.
Source: digitaleu, based on Eurostat 2025
Our approach may feel repetitive, but we really believe that you can’t do much as a country – or at the company, team, family level – if you don’t establish who is responsible for it, in general and now, in particular in the AI chapter, how they define their role, what vision they propose, and what key principles and objectives they put on the table to the leaders of the times for that national consensus I mentioned.
AI is probably the only major strategic technology for which Romania does not have a responsible institution. It is simply a governance vacuum on a crucial topic for the future.
Governance can start with an AI advisor at the Presidential Administration, where naturally the plans are longer-term, and can be doubled at the Executive level. It can mean a task force on emerging technologies that brings together bright minds from international academia, Romanian professors from renowned universities, and local leaders who can move quickly. It can go complementary with the GCIO office under the umbrella of the Deputy Prime Minister.
It can be a network to facilitate AI and interoperability, as Estonia tried with KrattAI a few years ago. It can be an inter-ministerial commission like in France, the Commission de l’intelligence artificielle that reports directly to the Prime Minister. It can be a public-private partnership like the U.S. AI Safety Institute (NIST) or a “center” like the Center for AI Standards and Innovation (CAISI), where technical standards are created, not just legal ones, and the developers of the strongest models make the results of safety tests available to the government. It can be a national program that brings together research, startups and government agencies under a single execution umbrella like AI Singapore (AISG).
It can be a local adaptation of an advanced model, such as the one in Ukraine: there, AI governance is based on the expansion of existing digitalization mechanisms, combining central coordination with clear responsibility of ministries for sectoral implementation, through already functional structures. This model also has elements similar to the GCIO network in Romania, which is, however, in an early stage in terms of developing interoperability at the government level. Across the border, the approach is phased: in a first phase, the emphasis is on coordination, experimentation and stimulating adoption through “soft law” instruments (sandboxes, codes of conduct, standards, public procurement), followed by a transition to full regulation and formal oversight mechanisms aligned with the AI Act as institutional capacity and the market mature.
It can be any model from the list above or another one that we haven’t mentioned. But it must be and then be a vision that aligns around a national consensus and a coherent action plan that defines, unites and coordinates the key actors of the ecosystem.
States that do not organize their AI will operate in the systems of others. Those that do not quickly choose a direction and hesitate in execution will permanently remain last, in another ranking with a clear negative impact on the well-being, education and safety of citizens.